June, 26, 2019

HPC System Administration Training

High Performance Computing System Administration Training - 4th to 8th March 2019

The training was organized for National Meteorological and Hydrological Services (NMHSs) system administrators and regional staff to build capacity on the usage and administration of the High Performance Computing (HPC) cluster installed at the centre and the modelling servers foreseen to be procured for NMHSs of SADC member states under the SARCIS-DR project. This activity is in line with the project implementation plan approved by the African Development Bank. The training was facilitated by CSC staff.

hpc trainingTraining Objectives
The objectives of the training were:
• to introduce National Meteorological and Hydrological Services to computer clusters and the concept of high performance computing, both the software and hardware aspects;
• to build capacity on the usage and administration of the HPC cluster installed at the CSC, and the server infrastructure being procured for member states
The required outcome of the training was for participants to:
• be able to develop and run parallel tasks on an HPC cluster;
• have the essential skills required to administer and maintain computer clusters and other Linux based early warning system infrastructure;

Date and Venue
The training was planned for a duration of 5 days, from 4th to 8th March 2019 at the CSC training room located in The Village, Gaborone, Botswana. The CSC has the early warning HPC system and training facilities required for the training.

Directors of National Meteorological and Hydrological Services were requested to nominate Information and Communication Technology (ICT) officers, with preference to be given to officers with prior experience in Linux and computer programming. The following nine (9) Member States managed to attend the training: Democratic Republic of Congo, Botswana, Eswatini, Lesotho, Malawi, Mozambique, Namibia, Zambia and Zimbabwe. Botswana, as the host country, nominate three officers to participate, hence the total number of participants from Member States was eleven (11).

Training Outcome

The training was a great success as indicated by the feedback from participants. The main concern from the participants is the duration of the training, with most feeling that one week is not sufficient to properly cover all the material. The objectives of the training were met, participants were equipped with the necessary skills to administer the HPC servers which are being procured for Member States. All participants were issued certificates of attendance at the end of the training. The Training Material is available for download.